Am Samstag, 17. Dezember 2016, 11:05:34 CET schrieb Gerlof Langeveld: > Hi Martin and Marc, > > On 12/14/2016 10:58 AM, Martin Steigerwald wrote: > > I didn´t post to LKML about the bug I reported in kernel upstream > > bugtracker > > yet. > > > > I ponder whether to report the second issue, the one about netlink, that > > Gerlof pointed out to the Debian bugtracker, since for Gerlof it does only > > happen with the Debian kernel and not with the CentOS kernel he tried and > > I > > also didn´t see this issue with my self-compiled kernel. > > Yesterday I did some further investigation about the second issue in > kernel 4.8.4: > > NETLINK command TASKSTATS_CMD_GET with nla_type > TASKSTATS_CMD_ATTR_REGISTER_CPUMASK results in error EINVAL (-22). > > Conclusion: it is a kernel bug. > Find my analysis below (I hope it is not too detailed :-). […]
I reported this as: Bug 190711 - Process accounting: Using the NETLINK inface, the command TASKSTATS_CMD_GET returns -EINVAL https://bugzilla.kernel.org/show_bug.cgi?id=190711 In addition with Bug 190271 - process accounting sometimes does not work https://bugzilla.kernel.org/show_bug.cgi?id=190271 both bugs are reported in Kernel bugtracker. I also added a note that Bug 180841 - Process accounting sometimes do not append records for terminated processes https://bugzilla.kernel.org/show_bug.cgi?id=180841 which refers to atop sometimes fails with a floating point exception or a trap exception https://bugs.debian.org/778598 my be solved at: https://bugzilla.kernel.org/show_bug.cgi?id=180841#c1 but I am not sure about that. Finally I am about to write a mail to LKML and kernel developers about both bugs. I will forward this one to the bug report (but not add it as Cc, as it may create a mail flooding to the bug report). Do you wish to be set on Cc in that mail? Thank you,
